Technical Consultant · Engineering Leader · Technical Architect · Staff Engineer
I build resilient systems and lead teams that ship.
Technical Consultant and Java technical expert with 8+ years building resilient banking and fintech platforms.
Learning on your own? Browse the courses →
Eight years ago I was at FlyTxt, chasing slow code paths in a telecom CRM platform, and one profiling pass cut latency 10x. Today I architect the Java and Spring Boot microservices behind the pricing and billing engine a bank’s deal desk runs on every day, 2M+ transactions at sub-second latency, and lead the engineers who build it. The through-line is one instinct: find what is actually broken and fix the real cause.
It is event-driven on Kafka and PostgreSQL, secured with OAuth2, Keycloak, and JWT, and deployed on Azure (AKS, ArgoCD, Terraform) and AWS (EC2, EKS, S3, SQS). At Way.com I moved authentication to OAuth2 and built idempotent Stripe payments. At Udaan a custom API gateway took uptime from 99.5 to 99.95 percent and cut latency 45 percent.
For 4+ years I have led teams of 10 to 30 engineers, including a Spring Boot upgrade across 20+ microservices that cleared 15+ CVEs. An architecture only holds if the team building it does too.
For about four years I have put AI to work inside that stack: an OpenAI-powered Pricing and Billing Copilot (Python, FastAPI, PyTorch) that halved L1 support resolution time and made me an AI Build Day 2026 finalist, and the RAG assistant on this site, built on a vector database with Hugging Face and NVIDIA models. Outside the day job I build end to end: this portfolio, and Temple Accounts and Billing, sold here as a downloadable product. If you are trying to get a system like these right, get in touch.
Open any area for what I've actually built with it, and the tools involved.
Built Way.com's API gateway for 5M+ daily requests at 99.95% availability, with Resilience4j, Redis, and Spring Cloud Gateway. At Zafin a Kafka-partitioned delta-processing ETL cut data-load time by 60%.
Built Zafins AI-powered Pricing and Billing Copilot: a PyTorch-based chatbot that self-trains on past queries and the answers that resolved them, cutting L1 support resolution time 50%. Also designed this sites own RAG pipeline, embedding its content with NVIDIA and Hugging Face models so its AI Assistant answers grounded in real site content.
Upgraded Spring Boot across 20+ microservices and eliminated 15+ CVEs. At FlyTxt, JProfiler and Hibernate tuning delivered a 10x latency reduction on the serving path.
Zafin's Corporate Pricing Engine prices and bills 2M+ transactions a day at sub-second latency. I was technical SPOC for Deal Manager's first production go-live at ING Bank, at 99.9% uptime on day one.
Mentored 10+ engineers at Zafin; SOLID and CQRS adoption cut defect leakage 40%. At Way.com I led 30+ features and pulled feature time-to-market from 6 weeks to 3.
Automated AWS/EKS deployments at SunTec, cutting manual effort 80%. At Udaan I ran the API gateway on Azure with Resilience4j, dropping P99 latency 35%.
Canary releases, A/B tests, and feature flags at Udaan cut deployment incidents 70% and doubled release velocity. Prometheus, Thanos, and Grafana brought MTTD from 15 minutes to under 3.
PostgreSQL is the system of record behind Zafin's event-driven pricing engine at 2M+ transactions a day. Redis caches Way.com's 5M+ daily-request gateway; Ignite Cache backs SunTec's streaming billing and payments.
Built Zafin's AI Pricing & Billing Copilot, cutting L1 resolution time by 50%. It sits on the live pricing and billing platform, not a side demo.
Shipped 30+ Way.com features and partner integrations — Amazon, getUpside, FuelRewards, Travelers, Cure — adding $1M+ incremental ARR. Client teams consume those through the REST and GraphQL APIs I designed.
Two ways to work with me
The patterns I use on real banking and fintech systems — as 1:1 help on your problem, or as courses you work through on your own.
Consultancy
Courses
Apps I've built and sell directly — a one-time purchase, no subscription.
Architecture decisions behind real production systems — event-driven design, resiliency, and the trade-offs that only show up at scale.
This site has its own AI Assistant, grounded in its own content through a real retrieval-augmented generation pipeline: embeddings, a relevance gate, access-aware retrieval, and a semantic answer cache. Here is how each piece earns its place.
How we cut L1 billing-support resolution time in half with a Python LLM service that streams answers over SSE and retrains itself on PyTorch from every query it has already resolved, sitting next to a Java and Spring Boot platform it never has to touch directly.
Architecting a corporate pricing and billing platform where the number has to be reproducible in front of an auditor a year later: why the pricing engine is a Kafka consumer instead of a service you call, how the command and query sides split, and what partitioning actually buys you.
No spam, no automated drip — just a note when something new is published.
